Flangers, delay time and modifying Boss flanger

[Kevin Mitchell]

Edited because I talk too much 

How would one double the clock speed of the BF-2 or HF-2? They are the same circuit but with two different BBDs (512 stage vs 1024 stage). I'd like to modify the BF-2 clock to be able to clock twice as fast at max setting to effectly give me the best of both pedals and quenching my hunger for an unreasonably wide sweep flanger.

Schematic here

Ideas... please give me ideas lol.

-KM

[ElectricDruid]

In theory, I *think* you'd change C30/47p to 22p to double the maximum clock speed, and then look at how to get the depth to go deeper so it could come back down to where it used to go.

But unfortunately I don't expect this to work. The MN3102 clock chip doesn't have enough oomph to drive the MN3207 at super-fast speeds. According to my notes the BF-2 uses 39KHz to 512KHz, so it's already pushing the chip way beyond what it's designed for. If that's right - I didn't make a note of where I found that information. You could 'scope the pedal you have, of course.
Also my own experiments with the MN3207 found that the output signal level started to drop significantly as you went above 500KHz, so the flanging effect would suffer even if you did manage to make the clock go faster.

Basically, if it was possible, I think Boss would have done it in the original pedal. Those Roland engineers weren't stupid, or even close.

[Mark Hammer]

Tom is correct.  Boss aimed for the highest clock rate they could squeeze out of the basic 3207/3102 combination, and left it at that.  They eventually produced the "Hi-Band Flanger, using an MN3204 chip, which, with only 512 stages, mimic what you're aiming for with your desired mod.  IF you could source an MN3204, you could just pop it in the socket and you'd have your wish.

The bottleneck in what you seek has two parts: 1) the high capacitance of the clock input pins on the Panasonic chips (roughly 700pf per 1024 stages), and 2) the low current-drive on the 3102 outputs. 

We are pretty well all familiar with the effects of cable capacitance and the virtues of buffering for preserving high end.  We know how a cap in the right/wrong place can add lag to a square wave.  The input capacitance of the BBD, and low current drive of the clock chip has the same sort of impact on the clock pulses arriving at the BBD, turning them from crisp trapezoidal pulses, achieving a relatively seamless handoff from stage to stage, to something much less crisp.  The bucket brigade turns into a bunch of people all saying "Just a sec while I hike up my pants here and pick my nose" before any buckets get passed.

That said, Panasonic MN30xx chips can be clocked MUCH higher than the 100khz spec'd on the datasheet.  I've witnessed one clocked up to 1.5mhz.  The solution is to buffer the MN3102 outputs, and provide some current drive.  This is achieved in many instances by paralleling a couple of invertor stages for each clock output, to provide both drive and buffering.  The A/DA Flanger does this, and so does the Hollis Ultraflange.

Is there room inside a BF-2 for some sort of daughterboard with the requisite add-on?  Couldn't tell you.  I imagine it depends on your skills.  Certainly a fabricated SMD board could fit, but one would still have to break the clock lines between MN3102 and MN3207, to run them to the daughterboard.

For chorus pedals using a 3207/3102 combo, it is possible to up the clock frequency and shorten the delay range by merely using a smaller-value cap for the 3102.  Sadly, one cannot apply that same general logic to pushing a BF-2 beyond its stock capabilities.  That said, there are specs, and there are tolerances.  Conceivably, the 47pf clock cap in your unit may well be a little larger in value than it says on the label, and the MN3102 may be able to push out slightly higher clock rates with comparative ease.  So I don't see anything terribly wrong, or bound for failure, with trying to titrate the clock-cap value down as low as is practically possible.  You may be able to push the min delay down below 1msec, though not likely down to 500usec or less, without the aforementioned external help.  However, 800usec-10.4msec would provide an audible improvement

Just in passing, I will note that the Feedback sound benefits and becomes less "box-ey" by dropping some of the bass via a reduction in the value of C7, from .047uf to something lower, like .022 or .01.  Since that also lowers the overall amplitude of the feedback, some tweaking of VR5 is probably permissible.

[Kevin Mitchell]

That's more than I could ask for. Thanks guys.

I certainly don't doubt that Roland knows their stuff! I'm just curious to know if we could push things a bit further. In the case of the HF-2, if we lower the clock speed we could actually get the results I'm after but sadly the BBD holding the signals for so long won't really sound so great. I could at least tinker and see what happens.

You know what... the Ibanez FL99 might do just what I want - or at least what Mark is suggesting. I always thought the intense switch just swapped some filtering (in retrospect I see how that's a waste of potential for a dual BBD). But I see the A output of the MN3210 feeds into input B! It also has the same clock design as the Boss flangers  So it's switching from 512 stages to 1024. The downside is both sections share the same clock so I won't be getting the full range desired in one sweep.
I actually have a completed FL99 clone and more MN3204 BBDs than I could home. Perhaps I'll rework the circuit into a 125B box 

Alternatively I could take something like the Utlra Flanger and have a crack at modifying the clock. Already have it half assembled on breadboard to play with. I could dodge the 3102 all together.

-KM

[Scruffie]

If you want a really wide sweep why don't you just build the A/DA Flanger? It doesn't really get any wider than that.

[Kevin Mitchell]

I would love an A/DA! I read it's as good as a flanger gets but opinions are biased. I appreciate you vouching on the sweep. I just haven't found a project I'm interested in for it.

Analogman claims that some reissues use MN3004, MN3210 or MN3204 but I can't find a schematic or image of a reissue PCB to confirm anywhere!

If anyone could dig up a schematic I'd gladly draw up a layout - not the SAD1024 or MN3007 versions.

-KM

[StephenGiles]

If you want a really wide sweep why don't you just build the A/DA Flanger? It doesn't really get any wider than that.

I agree, the ADA gives almost anything you would want from a flanger.............and more!

[Scruffie]

I would love an A/DA! I read it's as good as a flanger gets but opinions are biased. I appreciate you vouching on the sweep. I just haven't found a project I'm interested in for it.

Analogman claims that some reissues use MN3004, MN3210 or MN3204 but I can't find a schematic or image of a reissue PCB to confirm anywhere!

If anyone could dig up a schematic I'd gladly draw up a layout - not the SAD1024 or MN3007 versions.

-KM

Ahh, I see, I didn't realise the chip type was a prerequisite.

Well there's a schematic for an MN3010 version, adapting that to 2 x MN3004 would be pretty simple, with some TTL logic that could keep the 15V and become 2 x MN3204 or 1 x MN3210 (although that chip is practically extinct) or run at 9V with some adjustments.

[Mark Hammer]

Many views  about flangers are filtered through the sort of effect one is aiming for.  If slow dramatic "jet plane" is the objective, then three things matter: the minimum delay time, the sweep "ratio" between shortest and longest delay, the shape of the sweep.  Ideally, you want a minimum delay time somewhere sufficiently below 1msec, a sweep ratio of 30:1 or better, and a "hypertriangular" sweep that decelerates as it goes lower (i.e., as delay time lengthens) and then speeds up as it sweeps higher.  The BF-2 has a 13:1 sweep ratio (13msec down to 1msec), and uses a triangular LFO.  I tried to find out what the delay range (hence ratio) is for the FL99, but no luck.  The FL9, which uses an MN3207, claims a delay range of 1-12.8msec, so very close to the BF-2.  One might speculate that the FL99, when run in the non-intense mode (i.e., 512 stages), is 500usec-6.4msec, but that's speculation.  Either way, the sweep ratio is pretty modest.  The PAiA Hyperflange will do 150usec-15msec, or 88:1, courtesy of the CEM3340 used for the clock.

The FL99 appears to do some voodoo on the LFO output, and from your 2016 Youtube it seems to slow the sweep down as delay gets longer.  The Filter appears to simply invert the delay signal prior to mixing and feedback, with a wee bit of gain to compensate for the level change when using inverted signal to subtract from clean.

[Kevin Mitchell]

Ahh, I see, I didn't realise the chip type was a prerequisite.

Well there's a schematic for an MN3010 version, adapting that to 2 x MN3004 would be pretty simple, with some TTL logic that could keep the 15V and become 2 x MN3204 or 1 x MN3210 (although that chip is practically extinct) or run at 9V with some adjustments.

More of a preference really. Once I had seen that there were versions around with MN3010 or even better for me the MN3210/3204x2 I figured I could have a crack at homing a couple from my stash.

I found the schematic! It was in the ADA archive file from moosapotamus. Awesome 

Mark;
That's a great deal of info I appreciate it! I'm glad someone had already dug into these designs.
Now I'm curious about the PAiA - only because I may have a 3340 kickin' around. Maybe for the next flanger itch 


-KM

[Kevin Mitchell]

I guess the 3204 version is the ADA PBF Flanger.
All I should need to do is clone the original (mn3010 version), add a LT1054 pump + 15v regulator for power and supply each 3204 with the 9v supply.

Game on.

-KM

[Scruffie]

I guess the 3204 version is the ADA PBF Flanger.
All I should need to do is clone the original (mn3010 version), add a LT1054 pump + 15v regulator for power and supply each 3204 with the 9v supply.

Game on.

-KM

Not quite, as I said, you'd need TTL logic, the BBD expects a clock at half its supply voltage, so you can't put a clock run at 15V in to a BBD run at 9V without some way of converting it. There are chips to perform this job though.

There were some big box MN3004, MN3210 & MN3204 ADA flanger reissues, but I don't think anybody drew a schematic. I saw a gut shot and there was definitely quite a few changes but what they were...

[Rob Strand]

How would one double the clock speed of the BF-2 or HF-2? They are the same circuit but with two different BBDs (512 stage vs 1024 stage). I'd like to modify the BF-2 clock to be able to clock twice as fast at max setting to effectly give me the best of both pedals and quenching my hunger for an unreasonably wide sweep flanger.

FWIW, there was also a BF-2B Bass Flanger, one of those bronze-box era pedals.  It was a BF2 with a 512 stage chip no other changes from what i remember**.   (ElectricDruid covered the mod issues already.)


---------
EDIT:  ** Actually the high-pass cut-off frequencies on the delay and feedback might have been raised.
https://www.diystompboxes.com/smfforum/index.php?topic=81498.0

[Mark Hammer]

Kevin, go here: https://hammer.ampage.org/files/Hyperflange.PDF

If you're interested and desperate, I can spot you the PCB layout.

[Kevin Mitchell]

I guess the 3204 version is the ADA PBF Flanger.
All I should need to do is clone the original (mn3010 version), add a LT1054 pump + 15v regulator for power and supply each 3204 with the 9v supply.

Game on.

-KM

Not quite, as I said, you'd need TTL logic, the BBD expects a clock at half its supply voltage, so you can't put a clock run at 15V in to a BBD run at 9V without some way of converting it. There are chips to perform this job though.

Hmm... Do you think a pair of 2N7000 could work here?

-KM

[12Bass]

The sweep on my SAD1024 A/DA flanger clone is so wide that it is not very musically usable... that is unless you narrow the sweep somewhat.  Put another way, it isn't what I would consider a simple pedal to dial in due to the extreme range of adjustment.  That flexibility can create all sorts of weird and obnoxious noises as well as excellent chorus and flanging sounds; but getting those sweet tones requires careful adjustment from a skilled operator.  Shortly after I built it I lent it to a few friends and they had some difficulty dialing it in and did not seem that impressed.  IMO, even though it is a highly capable design, it is probably not a good choice for someone who wants a simple stomp-on-it swooshy flanger.  That said, if you know how to dial it in, it can sound wonderful....

[Kevin Mitchell]

My goal is to have the ultimate jet flanger and not so much chorus since I already have a handful of pedals that play the role well - mostly my DC-2 clone.

Scruffie - isn't there a way to route the 4049 to output 5v (TTL Logic)? If I'm understanding the datasheet if Vcc is 5v it'll output 5v while the input can still handle up to 15v.

I mean it's straight forward an easy to simulate;


-KM

[ElectricDruid]

Now I'm curious about the PAiA - only because I may have a 3340 kickin' around. Maybe for the next flanger itch 

One thing worth pointing out about the 3340 (or any of the recent clones: AS3340, V3340) is that it has an exponential response to Frequency CV, so that negates the need for hyper triangular waveforms for modulation*.
A 3340 clock will double in frequency for each X volts (1V/Oct typically, but not necessarily).

HTH,
Tom

*I don't think the maths is actually quite that simple, but it's at least as close as a distorted triangle will get you.

[Scruffie]

My goal is to have the ultimate jet flanger and not so much chorus since I already have a handful of pedals that play the role well - mostly my DC-2 clone.

Scruffie - isn't there a way to route the 4049 to output 5v (TTL Logic)? If I'm understanding the datasheet if Vcc is 5v it'll output 5v while the input can still handle up to 15v.

I mean it's straight forward an easy to simulate;


-KM

*Taps nose*

But thinking about it further... that might not actually be needed, you can probably just run the existing clock circuitry off 9V along with the BBD and then leave all the rest at 15V with some tweaking.

[Mark Hammer]

I like to describe the ideal "jet flanger" as - you'll pardon the expression - gradually "infecting" the signal.  It has this perceptual effect by starting the sweep well above the range where one would hear any comb-filtering.  And then as the delay-time is increased, the spectrum acquires more and more notches.  This stands in contrast to both chorus and phasers, in which there are always notches of some sort/quantity present.

There are two ways to accomplish that "infection".  One is to lowpass-filter the wet path, such that the resulting bandwidth is well below where any notches are created and audible.  The other way, using maximum possible wet-signal bandwidth, is to aim for as short a minimum delay as possible.  That's a big part of what made tape-flanging sound so awesome.  One minute, you're listening to the full bandwidth mix, and the next minute, this....thing....is slowly happening to it.

I might note that two of the most desirable flangers are the early Electric Mistress units, and the A/DA.    In addition to being able to nail very short delay times, and using optimal sweep waveforms, both of these use lowpass filtering on the wet path VERY sparingly.  Of course, the shorter the delay time, the more stingy one can afford to be with the lowpass filtering.

Take a gander at the difference in extent of lowpass filtering in the EM and BF-2.  I count 5 poles of LPF in the BF-2